The total instance price reflects the cost of the entire machine, while the per-GPU price shows how that cost relates to each GPU. Use the total rate for budgeting a full instance and the per-GPU rate for comparing GPU cost efficiency.
Use the deterministic table to verify the GPU count, exact instance type, and region for each listed configuration. These details provide the context needed to compare rates without implying current capacity or availability.
A lower per-GPU rate observed from another provider does not represent the same product as a AWS instance. Differences in configuration, infrastructure, service terms, and pricing scope limit direct comparison.
Limitations
Catalogue prices and regions are recorded observations. They do not prove live capacity, which may vary independently of the listed information.